ARTICLE

Received 16 Feb 2016 | Accepted 1 Dec 2016 | Published 27 Jan 2017

Highly tensile-strained Ge/InAlAs nanocompositesDaehwan Jung1, Joseph Faucher1, Samik Mukherjee2, Austin Akey3, Daniel J. Ironside4, Matthew Cabral5,

Xiahan Sang5, James Lebeau5, Seth R. Bank4, Tonio Buonassisi3, Oussama Moutanabbir2 & Minjoo Larry Lee1,6

Self-assembled nanocomposites have been extensively investigated due to the novel

properties that can emerge when multiple material phases are combined. Growth of epitaxial

nanocomposites using lattice-mismatched constituents also enables strain-engineering,

which can be used to further enhance material properties. Here, we report self-assembled

growth of highly tensile-strained Ge/In0.52Al0.48As (InAlAs) nanocomposites by using

spontaneous phase separation. Transmission electron microscopy shows a high density of

single-crystalline germanium nanostructures coherently embedded in InAlAs without

extended defects, and Raman spectroscopy reveals a 3.8% biaxial tensile strain in the

germanium nanostructures. We also show that the strain in the germanium nanostructures

can be tuned to 5.3% by altering the lattice constant of the matrix material, illustrating

the versatility of epitaxial nanocomposites for strain engineering. Photoluminescence and

electroluminescence results are then discussed to illustrate the potential for realizing devices

based on this nanocomposite material.

DOI: 10.1038/ncomms14204 OPEN

1 Department of Electrical Engineering, Yale University, New Haven, Connecticut 06511, USA. 2 Department of Engineering Physics, Ecole Polytechnique deMontreal, Montreal, Quebec, Canada H3C 3A7. 3 Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts02139, USA. 4 Department of Electrical and Computer Engineering, University of Texas at Austin, Austin, Texas 78758, USA. 5 Department of MaterialsScience and Engineering, North Carolina State University, Raleigh, North Carolina 27606, USA. 6 Department of Electrical and Computer Engineering,University of Illinois at Urbana-Champaign, Urbana, Illinois 61801, USA. Correspondence and requests for materials should be addressed to D.J.(email: daehwan.jung@engineering.ucsb.edu) or to M.L.L. (email: mllee@illinois.edu).

NATURE COMMUNICATIONS | 8:14204 | DOI: 10.1038/ncomms14204 | www.nature.com/naturecommunications 1

Germanium (Ge) is a semiconductor material with a longhistory of research behind it, including the first transistordemonstration1. One of the most intriguing aspects of Ge

is that strain can dramatically alter its band structure and enhanceits optical and electrical properties. Spurred by theoretical workshowing that tensile strain can convert Ge from an indirect-gap toa direct-gap semiconductor2, recent research has focused onapplying large biaxial or uniaxial tension using a range ofapproaches. For example, epitaxial growth of Ge thin films ontemplate layers with larger lattice constant (for example, InGaAsor GeSn) has enabled biaxial tensile strain up to 2.33% (refs 3,4).Top-down fabrication techniques have also been used to fabricateGe in biaxial or uniaxial tension using structures such asnanomembranes5, bridges6,7 and suspended nanowires8,9.

Self-assembled nanocomposites grown by spontaneousphase separation have been investigated in diverse materials,including oxide compounds10 and rare-earth monopnictide/III–Vsystems11,12. Furthermore, oxide nanocomposites with differentlattice parameters enabling strain tuning and enhanced propertieshave been reported13,14. For semiconductors, only unstrainedGe/GaAs nanocomposites have been reported to date15.

Here, we demonstrate spontaneous phase separation duringmolecular beam epitaxy (MBE) growth as an alternative approachto forming highly tensile-strained Ge nanostructures, coherentlyembedded in an InAlAs matrix (that is, Ge/InAlAs nano-composites). While the mutual immiscibility of Ge with III–Vmaterials provides the driving force for phase separation16,changes in growth kinetics enable significant control overnanostructure morphology, from nanowires to nanosheets. Thetensile strain in the Ge nanostructures is confirmed by Ramanspectroscopy, and we further show that changing the InxAl1� xAs

matrix composition enables strain tuning up to 5.3%, which, toour knowledge, is the largest biaxial tension realized in Ge to date;the cross-over from indirect to direct is predicted at B2% biaxialtension17. We believe that the group-IV/III–V nanocompositesdemonstrated here constitute a new materials platform forinvestigation of basic aspects of phase-separated growth, as wellas offering the ability to create ultra-high strain states andproperties that are otherwise inaccessible through conventionalgrowth and processing.

ResultsStructural properties of Ge/InAlAs nanocomposite. Figure 1aschematically illustrates the Ge/InAlAs nanocomposite growthprocess, which was performed by codeposition of Ge, In, Al andAs2 over a range of growth temperatures, growth rates and Gefluxes (see ‘Methods’ section). Atom probe tomography (APT) ofa Ge/InAlAs nanocomposite layer with 3.6% Ge grown at 500 �Creveals vertically continuous, Ge-rich columns in the [001]growth direction (Fig. 1b). The cross-sectional annular dark-field(ADF) scanning transmission electron microscopy (STEM) andcorresponding energy dispersive X-ray spectroscopy (EDX) Geelemental mapping illustrate that phase separation occursimmediately after the Ge atoms are incorporated into the filmwithout a two-dimensional wetting layer (Fig. 1c); a planarwetting layer is typically present in Stranski Krastanov growthmode of compressively strained nanostructures18. Planar-viewimages show that dark spots in the DF-STEM image preciselymatch the location of high Ge X-ray counts (Fig. 1d, seeSupplementary Fig. 1 for EDX maps of Ge, In, Al and As).The diameter of the nanowires is B5.5 nm with a density of1.8� 1010 cm� 2 in this sample. The nanowire densities can be

[001]b

e InAlAs overlayer

InAlAs buffer[001]

[110]

Ge nanowiresa

d[110]

GeDF-STEM

f

InAlAsmatrix

InAlAsmatrix

Ge nanowire

c[001]

[110]

GeDF-STEM

Ge In

InAlAs bufferInP substrate

Al As

Figure 1 | Structural characterization of Ge/InAlAs nanocomposite. (a) Schematic illustration of Ge/InAlAs nanocomposite growth.

(b) Three-dimensional APT image of Ge nanowires showing 10% Ge concentration iso-surface (red) and Al atoms (blue). Scale bar, 20 nm.

(c) Cross-sectional ADF-STEM and EDX images showing continuous columnar growth of Ge nanowires. Scale bar, 80 nm. (d) Planar-view

ADF-STEM and EDX Ge elemental mapping images. Scale bar, 60 nm. (e) Bright-field TEM image under g¼ (220) two-beam condition shows no evidence

of extended defects in a Ge/InAlAs nanocomposite layer sandwiched by InAlAs layers. Scale bar, 50 nm. (f) Cross-sectional Cs-corrected low-angle

ADF-STEM image reveals single-crystalline Ge nanowires with fully coherent interfaces. Scale bar, 4 nm. Images b,c were taken from samples with

3.6% Ge, growth temperature¼ 500 �C and images d–f were taken from a sample with 1.4% Ge, growth temperature¼ 520 �C.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14204

2 NATURE COMMUNICATIONS | 8:14204 | DOI: 10.1038/ncomms14204 | www.nature.com/naturecommunications

tuned by altering the Ge content in the nanocomposites(Supplementary Fig. 2 and Supplementary Note 1).

The bright-field TEM image in Fig. 1e shows coherent Genanowires without plastic strain relaxation, even with the highlattice mismatch of 3.72%. Furthermore, no extended defects suchas stacking faults, anti-phase domains or dislocations areobserved in either the nanocomposite layers or the Ge-freeInAlAs overlayers. In contrast, nanocomposites with much higherGe content did show plastic relaxation through dislocationformation due to the higher strain energy (Supplementary Fig. 3).Intense strain-induced contrast can be seen in the InAlAs matrixsurrounding the nanowires, which will be discussed more in alater section. The aberration-corrected low-angle ADF-STEMimage (Fig. 1f) shows a single-crystalline Ge nanowire andcoherent boundaries between the nanowire and InAlAs matrixmaterial with atomic resolution.

Growth kinetics of Ge/InAlAs nanocomposite. Unlike bulkphase separation at thermal equilibrium, spontaneous phaseseparation during MBE growth is surface-mediated, andtherefore adatom kinetics strongly affect the final microstructures.

The diagram in Fig. 2a illustrates the basic growth kinetics for thismaterials system, with a Ge/InAlAs metastable alloy regime and aphase-separated nanocomposite regime; all samples were grownat 1 mm per hour. We find that higher growth temperatures arerequired for lower Ge content films to form a phase-separatedstructure; the fact that phase separation occurs at all for sampleswith 1.4% Ge (upper left of diagram) is testament to thelow mutual solubility between Ge and InAlAs at 520 �C(Supplementary Fig. 4). Meanwhile, the lack of phase separationin 5.8–10.2% Ge samples grown at 460 �C shows that suppressingsurface diffusion ‘turns off’ phase separation, resulting inmetastable, single-phase layers (Supplementary Fig. 4). Ex situannealing of the metastable Ge/InAlAs samples (red regionof diagram) at 520 �C for 2 hours did not lead to any observablestructural changes by TEM, confirming that the formationof phase-separated Ge nanostructures relies on surfacediffusion12,19, as opposed to bulk diffusion20,21.

A decrease in growth rate leads to larger nanostructures with alower density, consistent with previously reported experimentalresults on other nanocomposite material systems (Fig. 2b)22.The insets of Fig. 2b reveal that we can tune the morphology ofthe Ge nanostructures from nanowires to nanosheets by changing

0.0 0.2 0.4 0.6 0.8 1.0

16

20

24

28

32

Nan

ostr

uctu

re s

paci

ng, R

(nm

)

Growth rate, � (µm h–1)

R ~ ( 1/� )0.293

b

dc

0 5 10 15 20 250

20

40

60

80

100

InAlAsmatrix

InAlAsmatrix

Ge nanowire

GeInAlAs

Ato

mic

per

cent

(%

)

Distance (nm)

2 4 6 8 10

460

480

500

520

Gro

wth

tem

pera

ture

(°C

)

Ge concentration (%)

a

Nanocomposite

Metastable alloy

Figure 2 | Effect of growth parameters. (a) Nanocomposite growth diagram showing effect of growth temperature and %Ge on phase separation. Red

circle denotes no observation of phase separation (Ge/InAlAs metastable alloy) and green circle denotes phase-separated growth (Ge/InAlAs

nanocomposite). All samples were grown at 1 mm per hour. (b) Measured nanostructure spacings versus growth rate. Black dash curve is a power law fit.

Insets show planar-view EDX Ge maps from 3.6% Ge samples grown at 1 mm per hour, 0.2 mm per hour, and 0.1mm per hour growth rate. Scale bars are

30 nm and arrow direction is [1�10]. (c) EDX linescans of Al, Ge, In, and As across nanowires. (d) APT Ge concentration iso-contour transect for nanowires.

Outermost iso-contour line (blue) represents 5% Ge and innermost iso-contour line (red) is 95% Ge for each nanowire. Scale bar, 10 nm.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14204 ARTICLE

NATURE COMMUNICATIONS | 8:14204 | DOI: 10.1038/ncomms14204 | www.nature.com/naturecommunications 3

the growth rate. The surface diffusion distance, r¼ffiffiffiffiffiffiffiffiffiffiffiffiDsd=u

p, is

known to affect the size of the phase-separated nanostructures,where Ds is the surface diffusivity, d is the monolayer thickness andu is growth rate23. As a result, the size of the Ge nanostructuresincreases with decreased growth rate, and the average spacing (R)between nanostructures is increased. Taking R to represent theminimum surface diffusion distance of Ge adatoms during growth,we found that r is proportional to (1/u)n with an exponent of0.293. Our experimentally measured exponent is smaller than thetheoretical value of 0.5 and the Monte Carlo simulation result of0.39 by Adams et al.23, but very close to the calculated exponent of0.27 by He et al.24. Thus, despite the novelty of Ge/InAlAsnanocomposites, classic descriptions of phase separation can beused to predict and control basic aspects of their growth.

EDX linescans in Fig. 2c reveal that the Ge concentration ofnanowires grown at 0.2 mm per hour peaks at B95% while the Ge

concentration is abruptly diminished to o1% in the matrix. TheAPT transect in Fig. 2d further confirms that the centre of eachnanowire is Z95% Ge and that the Ge concentration in thesurrounding matrix is o5% Ge (see Supplementary Fig. 5 forAPT transects of In, Al and As). In contrast, the Ge concentrationof nanowires grown at 1 mm per hour peaks at B75%(Supplementary Fig. 6 and Supplementary Note 1). Thus, whilethe growth diagram in Fig. 2a shows that phase separationand nanocomposite formation occur readily, EDX and APTdemonstrate that growth conditions encouraging longsurface diffusion distance (for example, low growth rate or highgrowth temperature) increase the extent to which phaseseparation takes place.

Strain properties of Ge/InAlAs nanocomposite. Raman spec-troscopy in Fig. 3a shows that a 3.6% Ge/InAlAs nanocompositesample (blue) has a Raman peak at 284.5 cm� 1, which is stronglyshifted from the bulk Ge (black) longitudinal optical (LO)phonon peak at 301.3 cm� 1. Care was taken to minimize anyeffects from sample heating, and while phonon confinement maycontribute to the observed shift25, biaxial tensile strain is thedominant cause (Supplementary Figs 7 and 8 and SupplementaryNote 2). We deduced the degree of strain in the Genanostructures using the equation, Do¼ b � biaxial tension,where b is a phonon strain shift coefficient; a value ofb¼ � 440 cm� 1 is adopted here26. The calculated tensilestrain is 3.8%, which is extremely close to the lattice mismatchbetween Ge and InAlAs (3.72%), and consistent with the Genanostructures being under biaxial tension near the surface.Surface depressions in the atomic force microscopy (AFM) image(Fig. 3b) matching the shape and size of Ge nanostructuresobserved in TEM are also consistent with the out-of-plane elasticrelaxation expected from biaxially tensile-strained structures at afree surface27. For comparison, the 3.6% Ge/InAlAs metastablealloy sample grown at 400 �C (red in Fig. 3a) shows only theIn–As LO phonon peak at B230 cm� 1 and Al–As LO phononpeak at B360 cm� 1 (Supplementary Fig. 9), but no additionalpeak near 284.5 cm� 1. It is important to note that Ramanspectroscopy analyzes only the top portion of the Genanostructures, because the penetration depth of the 633 nmlaser is only B30 nm in Ge. On the basis of the observed lack ofdislocations at the Ge/InAlAs interfaces, we speculate that the Genanostructures may take on a triaxial strain state further beneaththe surface, where elastic relaxation is not possible, but a detailedanalysis of strain versus depth is left for future work.

A further shifted Ge LO peak (magenta in Fig. 3a) is observedfrom Ge nanostructures embedded in an In0.76Al0.24As matrix,which demonstrates the ability to control the strain in theGe nanostructures (Supplementary Figs 10 and 11 andSupplementary Note 3 for more details regarding the InAlAsgraded buffer). The observed 23.3 cm� 1 shift to 278.0 cm� 1

corresponds to a 5.3% biaxial tensile strain, which is again, veryclose to the lattice mismatch between the Ge and In0.76Al0.24Asmatrix (5.15%); for comparison, Raman shift values of5.7–9.7 cm� 1 have been reported in earlier work on biaxialtensile-strained Ge3,28. Our results demonstrate that the tensilestrain in phase-separated Ge nanostructures is readily tunable tomuch higher values by changing the matrix lattice constant.

Raman polar patterns and optical properties. The Ge nanosheetsample displays strongly polar Raman intensity patterns followingIRamanBcos4(y) (Fig. 4a). Previous studies reported that Ramanscattering in semiconductor nanowires with diameter of B50 nmcan be dominated by a nearly perfect optical antenna effect,which is strong enough to mask the well-established bulk Raman

240 260 280 300 320

Inte

nsity

(a.

u.)

Raman shift (cm–1)

284.5 cm–1

Ge/In0.52Al0.48As

metastable alloy

Ge/In0.76Al0.24As

nanocomposite

Ge/In0.52Al0.48As

nanocomposite

Ge wafer

301.3 cm–1

278.0 cm–1

a

b

4.6 nm

0 nm

[110

]

Figure 3 | Raman spectroscopy and AFM image. (a) Raman spectra of Ge

wafer (black), Ge/In0.52Al0.48As metastable alloy (red), Ge/In0.52Al0.48As

nanocomposite (blue), and Ge/In0.76Al0.24As nanocomposite (magenta)

taken via 633 nm laser. Clear Raman shifts of Ge-Ge peaks are observed,

showing tensile strain in the Ge nanostructures. (b) 250� 250 nm2 AFM

image of Ge nanosheet sample showing surface depressions corresponding

to the size and density observed in TEM. Scale bar, 50 nm.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14204

4 NATURE COMMUNICATIONS | 8:14204 | DOI: 10.1038/ncomms14204 | www.nature.com/naturecommunications

selection rules29,30. Because the Ge nanosheets possess strongstructural anisotropy, the Raman intensity exhibits a cleardependence on the angle (y) between the incident laserpolarization direction and the Ge nanosheet orientation. Thefourth power dependence of Raman scattering intensity wasobserved previously in studies of single silicon nanowires, andexplained by a resonant enhancement of both the initialexcitation light (532 nm) and the Raman scattered light (Stokesshifted to 540 nm)30; such effects may also help to explain thefourth power dependence observed here. To the best of ourknowledge, this is the first observation of polarization-dependentRaman scattering intensity from semiconductor nanocomposites.Nanocomposites with embedded Ge nanowires (blue circles) donot follow the cos4(y) behaviour, and the effect of the bulk Ramanselection rules are still present; the Ge nanowire diameter may betoo small to exhibit strong optical antenna effects.

Figure 4b shows a broad photoluminescence (PL) signal peakedat B1,230 nm from the 3.6% Ge/InAlAs nanocomposite at roomtemperature. In contrast, a Ge/InAlAs metastable alloy thatcontains the same 3.6% Ge has no PL response in this wavelengthrange, indicating that the B1.0 eV PL peak is from the phase-separated Ge nanostructures; the bulk bandgap of InAlAs isB1.53 eV. The theoretically calculated direct bandgap of 3.75%biaxial tensile-strained Ge is B0.49 eV from Gamma (G) toheavy-hole (HH) and B0.12 eV from G to light-hole (LH)17. Inour surface-emission configuration, only G–HH emission can becollected, because the light emitted from G–LH recombinationpropagates in the in-plane direction5. The large blue-shift from0.49 to 1.0 eV likely results from strong quantum confinementeffects and unintentional doping, while the high full-width athalf-maximum indicates inhomogeneous broadening. However,the possible role of the non-uniform strain state of thenanostructures along with the different stoichiometry at theinterfaces between Ge and InAlAs also needs to be investigated31.

We also performed temperature-dependent PL andlow-temperature (T¼ 77 K) excitation-dependent PL on the3.6% Ge/InAlAs nanocomposite (Supplementary Fig. 12 andSupplementary Note 4). The integrated PL intensity increases andthe PL peak blue-shifts as the temperature decreases, which isconsistent with direct bandgap radiative recombination3,32.Excitation-dependent PL revealed that the PL intensities varylinearly with the excitation density. The inset of Fig. 4b showsthat annealing up to 600 �C can improve the PL intensity byB2� , which may result from a reduced density of point defectsin and around the Ge nanostructures33. Further study of opticalproperties such as time-resolved PL31 and photoacoustic

spectroscopy34 will be necessary to confirm the band structureof the highly tensile-strained Ge/InAlAs nanocomposite.

We grew and fabricated light-emitting diodes (LEDs) using aGe/InAlAs nanocomposite active region to investigate itspotential for device applications. The I–V curve in Fig. 4c showsthat both the nanocomposite and control device act aswell-behaved diodes with good rectification and low leakagecurrent, showing that the Ge nanostructures do not serve assignificant shunt paths. Extracted ideality factors for theGe/InAlAs nanocomposite LEDs and InAlAs diodes were B2due to a mixture of recombination in the depletion andquasi-neutral regions. The inset of Fig. 4c shows an electro-luminescence (EL) peak at l¼ 1,170 nm at room temperature,about 50 meV blue-shifted from the PL emission peak. Theintegrated EL intensity increases approximately linearly as thecurrent density increases from 22 to 46 A cm� 2. These initialdevice results demonstrate the possibility of integrating highlytensile-strained Ge/InAlAs nanocomposites into p–n junctiondevices such as light emitters and detectors.

DiscussionIn summary, we demonstrated the use of surface-mediatedphase separation to realize highly tensile-strained Ge/InAlAsnanocomposites. TEM showed coherently strained, single-crystalline Ge/InAlAs nanocomposites without observableextended defects, while Raman spectroscopy revealed 5.3% biaxialtensile strain, to our knowledge the highest value reported to datein Ge nanostructures. Ge/InAlAs nanocomposites are a new classof material allowing study of the electronic and optical propertiesof tensile-strained Ge, and the growth method may be extendableto other strained, semiconductor-based nanostructures for futureoptoelectronic applications.

MethodsMolecular beam epitaxy growth. Growth was conducted using a Veeco ModularGen-II MBE system. A 200 nm lattice-matched InAlAs buffer was first grown on(001) InP at 500 �C. Next, 300 nm nanocomposite layers consisting of tensile Genanostructures embedded in InAlAs were grown by codeposition of Ge, In, Al andAs2 over a range of substrate temperatures, growth rates and Ge fluxes. The V/IIIratio for the Ge/InAlAs film was kept at B25, and reflection high-energy electrondiffraction patterns were streaky throughout the growth.

Structural characterization via transmission electron microscopy. TEMsamples were thinned by standard mechanical grinding and additional Arion-milling. ADF-STEM (camera length 220 mm) and EDX elemental mappingand linescans were performed on a Tecnai Osiris microscope at an operatingvoltage of 200 kV. Low-angle ADF-STEM was performed using a probe corrected

800 1,000 1,200 1,400 1,600 1,800 2,000

PL

inte

nsity

(a.

u.)

Wavelength (nm)

NanocompositeMetastable alloy

Annealingb

–1.0 –0.5 0.0 0.5 1.0 1.5

0

10

20

30

40

50

Cur

rent

den

sity

(A

cm

–2)

Voltage (V)

NanocompositeGe-free InAlAs

1,000 1,100 1,200 1,300 1,400

EL

inte

nsity

(a.

u.)

Wavelength (nm)

Current

c

0.0

22.5

45.0

67.590.0

112.5

135.0

157.5

180.0

NanowiresNanosheetscos4(�)

a

[001]

[110]

Ei

�

[110]

Ge nanosheetsin InAIAs matrix

Figure 4 | Raman scattering and luminescence. (a) Polar (y) plots of Ge LO Raman peaks from Ge nanosheets (black circles) and nanowires (blue circles).

The Raman intensities are normalized. Inset illustrates that y is the angle between polarization of incident laser (Ei) and [1�10] direction of the sample. (b)

Room-temperature PL spectra from Ge/InAlAs nanocomposite (black) and from Ge/InAlAs alloy (red). Inset shows annealing effect of nanocomposite PL

intensity at room-temperature. (c) J–V curves from nanocomposite p-n diode (black) and Ge-free InAlAs diode (red). Inset shows EL spectrum from the

nanocomposite diode at room-temperature.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14204 ARTICLE

NATURE COMMUNICATIONS | 8:14204 | DOI: 10.1038/ncomms14204 | www.nature.com/naturecommunications 5

FEI Titan G2 60–300 kV operated at 200 kV with a beam current of B50 pA. Thecollection inner semi-angle and probe semi-convergence angle were B28 mrad and21 mrad, respectively. Atomic resolution STEM images were acquired with therevolving STEM (RevSTEM) method to eliminate drift distortion35. EachRevSTEM data set contained eight image frames of size 2,048� 2,048 pixels with a90� rotation between each successive frame. The time-dependent drift vector wasmeasured from the rotating frames, which were then corrected, aligned andaveraged to yield the final RevSTEM image.

Raman spectroscopy. Backscattering micro-Raman spectroscopy was carried outin ambient air in a Renishaw InVia RM 3,000 set-up, using a He/Ne 633 nmcontinuous-wave laser. A 100� Olympus objective (spot size at laser waistB1.0 mm) was used to focus the laser onto the samples, and the backscattered lightwas diffracted by an 1,800 lines per mm grating onto a liquid nitrogen-cooled CCDcamera. The incident laser power density of 1.72 mW mm� 2, as measured by aminiature power metre, was kept constant throughout the experiment. The datawas collected from three different spots on each sample and averaged.

Atom probe tomography. APT specimens were prepared by standard focused ionbeam (FIB) lift-out procedures. Using a Zeiss NVision 40 dual beam SEM/FIB(Gaþ ) equipped with an Omniprobe micromanipulator, sections of the Ge/InAlAsnanocomposite were attached to prefabricated micro tip coupons. Individualneedle-shaped APT specimens were then fabricated using FIB annular milling at anaccelerating voltage of 30 keV, followed by final sharpening and cleanup with anaccelerating voltage of 5 kV. Finished specimens had a tip radius-of-curvature ofB30 nm, and a shank half-angle of B18 degrees, as measured by SEM. APTwas performed using a Cameca LEAP 4,000 HR equipped with a 355 nm laser;specimens were run with a base temperature of 40 K, laser energy of 5 pJ per pulse,repetition rate of 100 kHz and target detection rate of 1%. At these conditions,significant As clustering was observed, as is common in III–V materials. APT datasets were then reconstructed and analysed using Cameca’s integrated visualizationand analysis software.

Photoluminescence/electroluminescence. Samples were optically pumped usinga 532 nm laser (photon energy significantly greater than the bandgap of InAlAsmatrix and InP substrate), and optical emission was collected by a spectrometerwith a liquid nitrogen-cooled InSb detector. The LEDs were tested by DC currentinjection and the light was collected by an InGaAs detector cooled to 253 K.

Light-emitting diode fabrication. We grew and fabricated p–n LED using a 3.6%Ge/InAlAs nanocomposite active layer (Supplementary Fig. 13a). We found that allof the nanocomposite layers are 1–3� 1018 cm� 3 n-type using room-temperatureHall effect measurements. Be-doped InAlAs was grown on top of the nano-composite layers to create p-n junction diodes. The LEDs were fabricated bydefining 1� 1 mm2 mesas, and gridded front metal contacts (Ti/Au¼ 20 nm/200 nm) and planar back metal contacts (AuGe/Ni/Au¼ 100 nm/20 nm/100 nm)were deposited for the anode and cathode, respectively. After front metal lift-offand mesa definition, the p-type InGaAs contact layer was etched using citric acidand hydrogen peroxide to improve light extraction. Devices were tested under DCcurrent injection at room-temperature, and the light was collected by an InGaAsdetector cooled to 253 K. Supplementary Fig. 13b shows J–V curves from the p-nnanocomposite LED and Ge-free InAlAs p–n junction in a semi-log plot.

Data availability. The data that support the findings of this study are availablefrom the corresponding authors on request.

References1. Bardeen, J. & Brattain, W. H. The transistor, a semi-conductor triode. Phys.

Rev. 74, 230–231 (1948).2. Fischetti, M. V. & Laux, S. E. Band structure, deformation potentals, and carrier

mobility in strained Si, Ge, and SiGe alloys. J. Appl. Phys. 80, 2234–2252 (1996).3. Huo, Y. J. et al. Strong enhancement of direct transition photoluminescence

with highly tensile-strained Ge grown by molecular beam epitaxy. Appl. Phys.Lett. 98, 011111 (2011).

4. Wirths, S. et al. Band engineering and growth of tensile strained Ge/(Si)GeSnheterostructures for tunnel field effect transistors. Appl. Phys. Lett. 102, 192103(2013).

5. Boztug, C., Sanchez-Perez, J. R., Cavallo, F., Lagally, M. G. & Paiella, R.Strained-germanium nanostructures for infrared photonics. ACS Nano 8,3136–3151 (2014).

6. Jain, J. R. et al. A micromachining-based technology for enhancing germaniumlight emission via tensile strain. Nat. Photonics 6, 398–405 (2012).

7. Suess, M. J. et al. Analysis of enhanced light emission from highly strainedgermanium microbridges. Nat. Photonics 7, 467–473 (2013).

8. Guilloy, K. et al. Tensile strained germanium nanowires measured by photocurrentspectroscopy and X-ray microdiffraction. Nano Lett. 15, 2429–2433 (2015).

9. Greil, J., Lugstein, A., Zeiner, C., Strasser, G. & Bertagnolli, E. Tuning theelectro-optical properties of germanium nanowires by tensile strain. Nano Lett.12, 6230–6234 (2012).

10. Moshnyaga, V. et al. Structural phase transition at the percolation threshold inepitaxial (La0.7Ca0.3MnO3)1�X:(MgO)X nanocomposite films. Nat. Mater. 2,247–252 (2003).

11. Singer, K. E., Rutter, P., Peaker, A. R. & Wright, A. C. Self-organizing growthof erbium arsenide quantum dots and wires in gallium-arsenide bymolecular-beam epitaxy. Appl. Phys. Lett. 64, 707–709 (1994).

12. Kawasaki, J. K., Schultz, B. D., Lu, H., Gossard, A. C. & Pamstrom, C. J.Surface-mediated tunable self-assembly of single crystal semimetallicErSb/GaSb nanocomposite structures. Nano Lett. 13, 2895–2901 (2013).

13. MacManus-Driscoll, J. L. et al. Strain control and spontaneous phase orderingin vertical nanocomposite heteroepitaxial thin films. Nat. Mater. 7, 314–320(2008).

14. Harrington, S. A. et al. Thick lead-free ferroelectric films with high Curietemperatures through nanocomposite-induced strain. Nat. Nanotechnol. 6,491–495 (2011).

15. Norman, A. G. et al. Ge-related faceting and segregation during the growth ofmetastable (GaAs)1� x(Ge2)x alloy layers by metal-organic vapor-phase epitaxy.Appl. Phys. Lett. 74, 1382–1384 (1999).

16. Ansara, I. & Materials Science International Team (MSIT). As-Ga-Ge ternaryphase diagram evaluation. SpringerMaterials MSI Eureka. Available at http://materials.springer.com/msi/docs/sm_msi_r_10_010650_01 (1994).

17. El Kurdi, M., Fishman, G., Sauvage, S. & Boucaud, P. Band structure and opticalgain of tensile-strained germanium based on a 30 band k � p formalism. J. Appl.Phys. 107, 013710 (2010).

18. Leonard, D., Krishnamurthy, M., Reaves, C. M., Denbaars, S. P. & Petroff, P. M.Direct formation of quantum-sized dots from uniform coherent islands ofInGaAs on GaAs-surfaces. Appl. Phys. Lett. 63, 3203–3205 (1993).

19. Norman, A. & Olson, J. Nanostructures produced by phase-separation duringgrowth of (III-V)1� x(IV2)x alloys. US patent 7,229,498 B2 (2007).

20. Heiss, W. et al. Centrosymmetric PbTe/CdTe quantum dots coherentlyembedded by epitaxial precipitation. Appl. Phys. Lett. 88, 192109 (2006).

21. Qi, M. et al. Ge quantum dots encapsulated by AlAs grown by molecular beamepitaxy on GaAs without extended defects. Appl. Phys. Lett. 104, 073113 (2014).

22. Tian, Y. et al. Ultrahigh-density sub-10 nm nanowire array formation viasurface-controlled phase separation. Nano Lett. 14, 4328–4333 (2014).

23. Adams, C. D., Srolovitz, D. J. & Atzmon, M. Monte-Carlo simulation ofphase-separation during thin-film codeposition. J. Appl. Phys. 74, 1707–1715(1993).

24. He, J. H., Carosella, C. A., Hubler, G. K., Qadri, S. B. & Sprague, J. A. KineticMonte Carlo simulations of nanoscale compositional patterning during filmgrowth of phase-separated systems. Phys. Rev. B. 73, 235406 (2006).

25. Wang, X., Shakouri, A., Yu, B., Sun, X. H. & Meyyappan, M. Study of phononmodes in germanium nanowires. J. Appl. Phys. 102, 014304 (2007).

26. Pezzoli, F. et al. Phonon strain shift coefficients in Si1� xGex alloys. J. Appl.Phys. 103, 093521 (2008).

27. Lenz, A., Tournie, E., Schuppang, J., Dahne, M. & Eisele, H. Atomic structure oftensile-strained GaAs/GaSb (001) nanostructures. Appl. Phys. Lett. 102, 102105(2013).

28. Bai, Y., Lee, K. E., Cheng, C., Lee, M. L. & Fitzgerald, E. A. Growth of highlytensile-strained Ge on relaxed InxGa1� xAs by metalorganic chemical vapordeposition. J. Appl. Phys. 104, 084518 (2008).

29. Chen, G. et al. Optical antenna effect in semiconducting nanowires. Nano Lett.8, 1341–1346 (2008).

30. Lopez, F. J. et al. Diameter and polarization-dependent raman scatteringintensities of semiconductor nanowires. Nano Lett. 12, 2266–2271 (2012).

31. Pavarelli, N. et al. Optical emission of a strained direct-band-gap Ge quantumwell embedded inside InGaAs alloy layers. Phys. Rev. Lett. 110, 177404 (2013).

32. Manna, S., Katiyar, A., Aluguri, R. & Ray, S. K. Temperature dependentphotoluminescence and electroluminescence characteristics of core-shellGe-GeO2 nanowires. J. Phys. D Appl. Phys. 48, 215103 (2015).

33. Dashiell, M. W. et al. Photoluminescence of ultrasmall Ge quantum dotsgrown by molecular-beam epitaxy at low temperatures. Appl. Phys. Lett. 80,1279–1281 (2002).

34. Eaves, L., Vargas, H. & Williams, P. J. Intrinsic and deep-level photoacoustic-spectroscopy of GaAs (Cr) and of other bulk semiconductors. Appl. Phys. Lett.38, 768–770 (1981).

35. Sang, X. H. & LeBeau, J. M. Revolving scanning transmission electronmicroscopy: correcting sample drift distortion without prior knowledge.Ultramicroscopy 138, 28–35 (2014).

AcknowledgementsThis work was supported by NSF DMR 1506371 as well as a Dubinsky New ResearchInitiative Grant. Facilities use was supported by YINQE and NSF MRSEC DMR 1119826.The work at Ecole Polytechnique was supported by NSERC (Discovery Grant), CanadaResearch Chair, and Canada Foundation for Innovation. The work at MIT was supported

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14204

6 NATURE COMMUNICATIONS | 8:14204 | DOI: 10.1038/ncomms14204 | www.nature.com/naturecommunications

Bay Area Photovoltaic Consortium (BAPVC) under Contract No. DE-EE0004946.The work at UT-Austin was supported by a Multidisciplinary University ResearchInitiative from the Air Force Office of Scientific Research (AFOSR MURI AwardNo. FA9550-12-1-0488) and NSF DMR 1508603. The authors acknowledge the use of theAnalytical Instrumentation Facility (AIF) at North Carolina State University, which issupported by the State of North Carolina and the National Science Foundation.

Author contributionsD.J., J.F., and M.L.L. conceived the idea, and D.J. and M.L.L. designed the experiments.D.J. grew the nanocomposite materials and performed DF-STEM, EDX, TEM,XRD, AFM, and Raman spectroscopy via 532 nm laser. D.J. fabricated the LEDs andmeasured I-Vs. A.A. and T.B. conducted the APT on the nanocomposites. S.M. andO.M. performed Raman spectroscopy via 633 nm laser. M.C., X.S. and J.L. carried outCs-corrected low-angle ADF-STEM. D.J.I. and S.R.B. conducted PL and EL measure-ments. D.J. and M.L.L. wrote the paper with contributions from all the other authors.

Additional informationSupplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Competing financial interests: The authors declare no competing financialinterests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

How to cite this article: Jung, D. et al. Highly tensile-strained Ge/InAlAsnanocomposites. Nat. Commun. 8, 14204 doi: 10.1038/ncomms14204 (2017).

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

This work is licensed under a Creative Commons Attribution 4.0International License. The images or other third party material in this

article are included in the article’s Creative Commons license, unless indicated otherwisein the credit line; if the material is not included under the Creative Commons license,users will need to obtain permission from the license holder to reproduce the material.To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14204 ARTICLE

NATURE COMMUNICATIONS | 8:14204 | DOI: 10.1038/ncomms14204 | www.nature.com/naturecommunications 7